Highlights Client Stori Category Business / Risk Management Assessment System Engagement Solution Upgrade, New Feature Development Technologies Brief Stori is a digital platform focused on providing digital financial services to underserved customers. It is a leading financial service provider in Mexico with offices across the globe. One of the key services offered by the company is buying out loans from financial institutions like banks after conducting due diligence. The Challenge The key challenge with the existing solution was the excessive time taken to evaluate each loan due to dependency on manual processes. It usually would take 48hrs to categorize and evaluate loan risks. The client’s objective was to fast-track the loan evaluation process and save time and cost by enhancing the platform without affecting day-to-day operations. Limitations of Existing Solutions The existing solution had no capability of uploading transcripts and communication information between the borrower and the lender that were in different formats. The client had to manually review each loan’s transcript to know whether the loan can be classified as performing or non-performing. There were no automated scoring mechanisms. The client had difficulty viewing how the loans are divided to identify trends and make informed decisions. Proposed Solution After the initial exploration of the client’s issues and key objectives, team Galaxy recommended AI-based automation by using Machine learning and NLP (Natural Language Processing) integration to reduce the time taken to evaluate each loan. Text classification was employed to assign each transcript between borrower and lender to know how the loan is distributed across each category such as Bankruptcy, Foreclosure, Fraud or Disputes. Based on the categories, the system was able to assign risk ratings to loans. The project started as a 3-4 month engagement which led to 2+ years of association for ongoing upgrades. Key Deliverable Configuring Machine Learning and NLP Data Import Developing and Training AI model Reporting and Analytics Development Stages 1 Research & Exploration 2 UI/UX Design 3 Integrating Machine Learning & NLP Algorithms 4 Model Training 5 QA & Bug Fixing 6 Final Review with Stakeholders 7 Deployment of the AI model Key Features Enhancements Multi-User Environment We provided a multi-user environment with the ability for the admin to log in as an end-user. This feature empowered the admin to address user concerns, facilitate training, ensure system quality, and efficiently handle user-specific tasks within the loan management process. New Feature Development Data Import The data import functionality feature is used for importing digital versions of conversational transcripts of a borrower and the lender in the database. This feature simplifies the process of entering large volumes of data(CSV, Excel or XML format) in a structured & automated way to know whether the loan would be performing or non-performing. Text Classification The text classification feature is used to normalize the comments that have short forms, typos, and word segmentation. This feature classifies text into different categories that have been predefined in the system. Machine Learning & NLP Integration Before implementing this feature we integrated and trained an AI model to learn to read and process digital transcripts. With this feature, one can know the risk analysis of loans in just 5-8 seconds which earlier used to take 48hrs through manual processing. Loan Reporting Dashboard The dashboard provides a summary of loans with categories such as Bankruptcy, Foreclosure, Fraud, or disputes. The users who have access to the dashboard can view the risk score of the loan, create a custom report and download a pdf of the report or share it with the team. Delivery & Deployment Team Galaxy successfully developed the module and integrated it with the existing system. Customization and configuration were done to align the system with the client’s specific workflows. We provided training to the client’s team to leverage the system’s capabilities to their full potential. Following this, we also engaged with the client for maintenance and support. Value Creation & Impact of the Solution Commercial Impact The commercial impact of automating the loan management system with machine learning and NLP was being able to reduce loan processing time from an average of 48 hours to 5 – 8 seconds. This also helped to reduce operation costs and it also opened more opportunities for business as they were able to evaluate more loan applications in less time. Impact on Brand Equity With AI integration & automation, the system provides more accuracy and faster loan processing time. This has helped the client to strengthen the brand equity and become a trusted brand for banks to partner with. Platform for Future Innovation The system was modernized for the client keeping in mind that it can be customized as per different use cases in the future. The client can increase the capability with innovative features like automating document verification, offering personalized loan recommendations and more without changing the architecture of the development.